Search

CN-122025158-A - Method and system for monitoring cardiac function rehabilitation state of crowd after coronary heart disease operation

CN122025158ACN 122025158 ACN122025158 ACN 122025158ACN-122025158-A

Abstract

The heart function rehabilitation state monitoring method for the crowd after coronary heart disease operation comprises the steps of constructing a third-order dynamic baseline adapting to postoperative pathological evolution, stress-physiological coupling modeling based on a double AI cooperative framework, and trend early warning of equal condition longitudinal pairing. The heart function rehabilitation state monitoring system comprises an electrocardiosignal preprocessing unit, wherein the electrocardiosignal preprocessing unit is connected with a multidimensional electrocardio feature extraction unit, the multidimensional electrocardio feature extraction unit is respectively connected with a third-order individuation baseline management unit and a prepositive stress identification AI unit, and the third-order individuation baseline management unit and the prepositive stress identification AI unit are respectively connected with a core correlation deduction AI unit. The method and the system for monitoring the heart function rehabilitation state of the crowd after coronary heart disease operation thoroughly solve the problem of high false alarm of the traditional fixed baseline in early postoperative period by triple-stage design and double calibration mechanism by deeply binding the dynamic baseline construction with the pathological evolution law after coronary heart disease operation.

Inventors

  • YANG DINGDING
  • YANG BOXUAN
  • Yang Kangwu

Assignees

  • 昆明佰卓嘉晟经贸有限公司
  • 西安市鄠邑区品言二外教育科技有限公司

Dates

Publication Date
20260512
Application Date
20260413

Claims (10)

  1. 1. The heart function rehabilitation state monitoring method for the crowd after coronary heart disease operation is characterized by comprising the following steps of: step 1, constructing a third-order dynamic baseline adapting to postoperative pathological evolution; step 2, stress-physiological coupling modeling based on a double AI collaborative architecture; and 3, carrying out trend early warning on longitudinal pairing under equal conditions.
  2. 2. The method for monitoring cardiac functional recovery status of a population after coronary heart disease surgery according to claim 1, wherein the step 1 comprises: step 1.1, establishing an initial baseline of a cold start period; step 1.1.1, processing the original electrocardiographic data to obtain qualified data for constructing an initial baseline; The qualified data for constructing the initial base line simultaneously meets the following conditions that the signal quality SQI is more than or equal to 60 minutes, the continuous duration is more than or equal to 5 minutes, the heart rate fluctuation range is less than or equal to 10 times/minute, the heart rate fluctuation is free from severe fluctuation, the abnormal heart rate and the electrocardio rhythm are regular, and the patient is in a lying rest state during data acquisition, and is free from mood and movement interference; Step 1.1.2, extracting three major core electrocardio characteristics in qualified data, respectively calculating arithmetic average values of six core indexes of heart rate, SDNN, RMSSD, LF/HF, QRS wave width and R wave amplitude as initial baseline reference values, and simultaneously calculating +/-2 sigma confidence intervals of all indexes to form initial safety baseline envelope; The heart rate storage related features comprise time domain SDNN, RMSSD, pNN50 0 and MeanNN, frequency domain LF, HF, LF/HF, nonlinear SD1, SD2 and sample entropy, the stroke volume storage related features comprise QRS wave width, QRS wave width variation coefficient, R wave amplitude stability, ST-segment micro-offset, T wave electric alternation and T wave symmetry, and the recovery capability features comprise heart rate recovery delay and HRV rebound rate; step 1.2, constructing a baseline anchoring period stress-physiological response benchmark library: step 1.2.1, combining real life track data of a patient after discharge, analyzing long-time electrocardiographic data of the patient in real time through a prepositive stress identification AI unit, outputting stress or non-stress classification labels and confidence degrees of 0-1 interval, screening effective stress time periods and corresponding non-stress rest time periods, and completing paired data acquisition of stress state and electrocardiographic index response; the paired data simultaneously meets the following conditions that the signal quality SQI of a stress period and a rest period is more than or equal to 60 minutes, a high-confidence stress event, namely the confidence of a stress state is more than or equal to 0.7, the time difference between the stress period and the corresponding rest period is less than or equal to 2 hours, circadian rhythm interference is eliminated, and no data artifact exists; Step 1.2.2, extracting core features related to heart reserve functions under different stress loads, constructing a stress-physiological response reference library specific to a patient individual, defining normal response ranges under various stress scenes, forming a standardized scale for rehabilitation evaluation, completing individual light fine adjustment of a double AI model based on paired data, and adapting stress-physiological response rules of the patient individual; Core features related to heart reserve function under different stress loads include heart rate peak offset, HRV decline ratio, ST-segment micro-offset, and heart rate recovery delay; the stress-physiological response reference library is stored in a structured way, and each paired data unit comprises four core dimensions, namely a standardized stress load factor, a stress period core electrocardio characteristic value, a corresponding resting period core electrocardio characteristic value and a stress-physiological response amplitude; When constructing a stress-physiological response benchmark library specific to a patient individual, fitting normal response intervals of all features under different stress loads by using at least 20 groups of effective paired data, wherein the interval range is mean value +/-2 sigma, so as to form a complete individuation response benchmark library; step 1.3, self-adaptive calibration in a dynamic evolution period; Adopting a dual mechanism of incremental learning algorithm and medical node forced calibration, and combining an abnormal baseline drift interception mechanism to complete long-term dynamic adaptation of a baseline: The method comprises the steps of automatically extracting effective resting state data and stress-physiological paired data of a patient in the week every week, performing incremental fine adjustment on a baseline, enabling the single-week fine adjustment amplitude not to exceed a preset proportion, automatically extending to the next week to avoid baseline deviation caused by small sample data if the effective data amount in the week is insufficient, updating a confidence interval and a characteristic response interval of the baseline in the fine adjustment process, automatically triggering full baseline reset if the patient administration information adjustment is identified or a post-operation preset time node is reached, automatically updating or manually inputting a patient coronary angiography review result, a cardiac function examination report and administration adjustment information by a user, reconstructing a stress-physiological response reference library based on the latest clinical data and the effective monitoring data, and completing baseline full update; the abnormal baseline drift interception mechanism is used for automatically triggering abnormal verification if the single-cycle variation amplitude of the baseline of the patient is monitored to exceed a preset proportion, suspending baseline fine adjustment if the pathological abnormality is confirmed to exist, triggering risk early warning, and completing baseline calibration again after eliminating abnormal data if the physiological fluctuation or data interference exists.
  3. 3. The method for monitoring cardiac functional recovery status of a population after coronary heart disease surgery according to claim 2, wherein the incremental fine tuning of the baseline is performed as follows: Step B1, incremental update data access verification; Completing compliance screening of the effective data of the week, if the data of the week does not meet the admission condition, automatically extending to the next week, keeping the current baseline unchanged if the accumulated forward extending is not more than 2 weeks, and starting incremental updating after the data reach the standard; Compliance screening of valid data at this week requires that the following conditions be met simultaneously: The resting state data comprise signal quality SQI not less than 60 minutes, continuous duration not less than 5 minutes, heart rate fluctuation amplitude not less than 10 times/minute, no severe fluctuation, no abnormal heart rate and regular electrocardio rhythm, the patient is in a lying resting state during data acquisition, no emotion and movement interference are avoided, and a single-week effective resting state segment is not less than 10 segments; Stress-physiological paired data, wherein the signal quality SQI of a stress period and a resting period is more than or equal to 60 minutes, the confidence level of a stress state is more than or equal to 0.7, the time difference between the stress period and the corresponding resting period is less than or equal to 2 hours, no data artifact exists, and Shan Zhouyou effective paired data are not less than 15 groups; The data has no abnormal mark, namely no triggering of secondary and above risk early warning, no pathological characteristic abnormality, no medicine application/operation state change and no total reset requirement triggered by clinical review nodes; Step B2, incremental feature statistics and relative deviation calculation; for the effective data passing the admission check, after six core indexes, namely heart rate, SDNN, RMSSD, LF/HF, QRS wave width and R wave amplitude are extracted, statistics calculation and deviation quantization are respectively completed: calculating basic statistics, namely respectively calculating the mean value and standard deviation of each index in the effective resting state data of the week and the mean value and standard deviation of stress response amplitude of each feature in stress-physiological paired data; and (3) relative deviation quantification, namely calculating the relative deviation of the current baseline of the current week statistical value, wherein a single-feature relative deviation calculation formula is as follows: , In the formula, Is the first The relative deviation of the core indicators of the terms, For the effective data mean of the index at this week, The index core reference value determined for the anchor period; Calculating response interval deviation, namely calculating upper and lower limit deviation of stress response amplitude of each index in the week relative to the current response interval aiming at a corresponding stress-physiological response reference library; step B3, updating an incremental baseline with hard constraint; based on the characteristic relative deviation result, only performing incremental fine adjustment on the upper and lower limits of confidence interval and the upper and lower limits of stress-physiological response interval of each index, and core reference value The whole course is fixed, and the following amplitude constraint and updating rule are executed: hard amplitude constraint rule, that is, interval adjustment amplitude of single index and single circumference is relative to core reference value The average fine tuning amplitude of the six core indexes does not exceed the first preset proportion; Incremental updating weight distribution, namely adopting a sliding time window weighting mechanism, wherein the weight ratio of the newly added effective data in the week is less than that of the historical effective data from the anchor period to the previous week; The grading updating rule is that incremental fine adjustment of a corresponding interval is started only when the relative deviation of a single index is in a preset interval, the deviation is less than the lower limit of the preset interval, the original interval is kept unchanged, and when the deviation is more than or equal to the upper limit of the preset interval, an abnormal baseline drift interception mechanism is directly triggered and the baseline updating is not executed; The interval updating calculation mode is that the upper limit value calculation formula after updating the upper limit of the confidence interval is as follows: , In the formula, In order to update the upper bound of the post-interval, In order to update the upper bound of the pre-interval, In order to increase the learning rate, The lower limit update of the confidence interval adopts a symmetrical calculation logic with the upper limit update of the confidence interval; step B4, double checking and library dropping of the increment updating result; After the baseline fine tuning is completed, a double check mechanism is executed, and after the updated result meets the constraint requirement, the final library falling is completed: checking the trimming amplitudes of six core indexes one by one, and if the single index trimming amplitude exceeds the preset proportion, automatically carrying out cut-off correction according to the upper limit of the preset proportion; The logic consistency check comprises the steps of checking an updated baseline interval, ensuring that the problems of interval inversion, reference value deviation and response logic contradiction do not occur, ensuring that the anchoring relation between the updated baseline and the core reference value of an anchoring period is unchanged, storing the updated baseline confidence interval and the stress-physiological response interval into a third-order individuation baseline envelope after the checking is passed, synchronously updating a historical baseline version, keeping a complete updating track traceable, and directly discarding the updating result when the checking is not passed, so as to keep the original baseline unchanged.
  4. 4. The method for monitoring cardiac functional recovery status of a population after coronary heart disease surgery according to claim 1, wherein the step 2 comprises: 2.1, a pre-stress identification AI layer is adopted to analyze HRV full-dimension characteristics extracted from long-time electrocardiographic data of a patient in real time by adopting a pre-stress identification model, a stress or non-stress classification label and a confidence coefficient score of a 0-1 interval are output, and the integral of the confidence coefficient on a time axis is defined as a standardized stress load factor; The HRV full-dimension features comprise time domain features, frequency domain features and nonlinear features, wherein the time domain features comprise SDNN, RMSSD, pNN, meanNN, the frequency domain features comprise LF, HF and LF/HF, and the nonlinear features comprise SD1, SD2, SD1/SD2 and sample entropy; The mathematical expression of the normalized stress load factor is: , In the formula, The stress state confidence level (value 0-1) is output for the t moment model; In order to initiate the time of the stress event, Is the stress event termination time; SLF is a standardized stress load factor, the value is 0-1, and the larger the value is, the higher the stress load intensity is; Step 2.2, core correlation deduction AI layer, adopting a core correlation deduction model, taking a standardized stress load factor as core constraint, inputting three main types of core features directly related to heart reserve function, automatically excavating special feature correlation of an individual through an attention mechanism, outputting an individual dynamic feature weight matrix, namely automatically identifying a core index most related to the heart reserve function aiming at a current patient and distributing weight, constructing a response function from emotion pressure intensity to abnormal degree of the heart feature, calculating emotion-electrocardio response function slope K, and outputting a quantized CRI heart rehabilitation reserve index of 0-100 minutes through nonlinear fusion calculation; the core indexes most relevant to the heart reserve function comprise heart rate peak value offset, HRV full-dimension index, QRS wave width variation coefficient, R wave amplitude stability, ST segment micro offset, T waveform state characteristics, heart rate recovery delay and HRV rebound rate; The linear regression expression of the emotion-electrocardiographic response function slope K is: , Wherein X is a standardized stress load factor SLF, Y is an electrocardio characteristic abnormality degree comprehensive value, K is a response function slope, and B is a constant term; k value reflects the tolerance capacity of heart function of the patient to stress, the normal range is 0.1-0.5, K <0.3 is good in heart stress tolerance capacity, K <0.3 is reduced in heart stress tolerance capacity, K is more than or equal to 0.6, and the heart stress tolerance capacity is obviously insufficient; step 2.3, mapping the rehabilitation state, namely establishing a quantitative mapping relation between the CRI index and the heart reserve function; the CRI index is more than or equal to 80 minutes, the K is less than 0.3, the rehabilitation state is good, the heart reserve function is sufficient, the heart rate reserve and the stroke volume reserve are in the normal range, and no obvious rehabilitation decline risk exists; the CRI index is 60-79 minutes, K is more than or equal to 0.3 and less than or equal to 0.6, the rehabilitation state is stable, the heart reserve function is slightly reduced, the reserve function is basically normal, and the risk of continuous decline of the rehabilitation state needs to be alerted; The CRI index is less than 60 minutes, K is more than or equal to 0.6, the recovery state is declined, the heart reserve function is obviously insufficient, the reserve function is damaged, the recovery state is prompted to have continuous worsening risk, and the monitoring is required to be enhanced.
  5. 5. The method for monitoring cardiac functional recovery status of a population after coronary heart disease surgery according to claim 4, wherein the pre-stress recognition model structure comprises: An input layer for receiving HRV full-dimension characteristics output by the multi-dimension electrocardio characteristic extraction unit and finishing standardization and anomaly filtering of input data; The core reasoning layer adopts a random forest integrated learning algorithm as a basic framework and comprises the following steps: the bottom layer general feature extraction module consists of a pre-trained fixed weight decision tree and is responsible for learning general physiological association rules between stress states and HRV features, and weights are fixed in the whole course after model pre-training is finished and do not participate in subsequent individual fine adjustment; The upper-layer individual adaptation classification module is composed of decision trees capable of finely adjusting weights, light fine adjustment is completed based on proprietary data of the anchoring period of the patient after coronary heart disease operation, the primary recognition results output by the bottom-layer module are subjected to weighted fusion, and a final stress state comprehensive probability result is output; The output layer is used for synchronously outputting two core results, namely a stress/non-stress classification standardized label and a stress state confidence score in a 0-1 interval, wherein the two core results provide unique core input for the calculation of a follow-up standardized stress load factor and the modeling of stress-physiological coupling; the real-time analysis of the HRV full-dimension characteristics extracted from the long-time-range electrocardiographic data of the patient by adopting the pre-stress identification model comprises the following steps: Step C1, input characteristic standardization and abnormality filtration; Receiving HRV full-dimension characteristics output by a multi-dimension electrocardio characteristic extraction unit, adopting a zero-mean normalizer synchronously packaged in a basic model training stage to normalize all input characteristics, eliminating dimension differences of different characteristics, adopting a 3 sigma principle to complete characteristic outlier filtering, filling characteristic outliers exceeding a mean value +/-3 sigma range, and adopting patient individuation baseline median values of corresponding characteristics; step C2, screening high-discrimination core features; Based on the feature screening result which is completed through Cohen's d effect quantity analysis in the pre-training stage, extracting a core feature subset with high distinction degree corresponding to the stress state and the non-stress state from the standardized HRV full-dimension features, and eliminating redundant low distinction degree features; Step C3, extracting and reasoning general features of the bottom layer; Inputting the screened core features into a model bottom layer general feature extraction module, wherein the weight of the module is fixed, only general physiological characterization related to the stress state in the input features is extracted, and outputting a preliminary classification probability result of each decision tree in the module to the stress state, thereby completely retaining the general physiological rule of the stress-HRV features learned in the pre-training stage; Step C4, upper-layer individual adaptation classification fusion reasoning; Inputting the preliminary probability result of the single decision tree output by the bottom layer module into an upper layer individual adaptation classification module of the model, carrying out weighted fusion on the preliminary probability result output by the bottom layer, and finally outputting a stress state comprehensive probability value in a 0-1 interval; Step C5, outputting stress/non-stress classification labels; Based on a preset classification judgment rule, carrying out classification judgment on the output stress state comprehensive probability value, and outputting a standardized classification label, wherein when the comprehensive probability value is more than or equal to a classification threshold value, the stress label is output; step C6, outputting confidence scores of the intervals of 0-1; based on the consistency of classification results of all decision trees of a model core reasoning layer, calculating and outputting stress state confidence scores of a 0-1 interval, wherein a calculation formula is as follows: , In the formula, For the number of decision trees for which the classification result is consistent with the final classification label, The total number of decision trees for the model.
  6. 6. The method for monitoring cardiac functional recovery status of a population following coronary heart disease surgery according to claim 4, wherein the core correlation deduction model comprises: The input layer takes a standardized stress load factor as a core constraint, and synchronously inputs three main core characteristics, namely heart rate reserve characteristics comprise heart rate peak value offset, SDNN, RMSSD, LF/HF and pNN50, stroke volume reserve associated morphological characteristics comprise QRS wave width, QRS wave width variation coefficient, R wave amplitude stability, ST-segment micro offset and T wave symmetry, and recovery capability characteristics comprise heart rate recovery delay and HRV rebound rate; The feature coding layer is used for independently coding the modal features by adopting two full-connection layers, mapping the original features into high-dimensional feature characterization, adopting a ReLU activation function for each layer, and adding a Dropout layer; The attention weighting layer is used for calculating the dynamic weight of each feature through a self-attention mechanism, wherein the weight calculation is strongly related to the standardized stress load factor of a patient; The nonlinear fusion layer and the output layer are formed by splicing the weighted attention characteristics, completing nonlinear fusion through two layers of full-connection layers, and finally passing through Activating function to output recovery score in 0-1 interval, multiplying 100 to obtain CRI heart recovery reserve index of 0-100 score, and fitting based on linear regression to output emotion-electrocardio response function slope ; The asymmetric loss function formula is: , Wherein, the As a function of the loss of asymmetry, To overestimate the prediction error of the heart reserve function, Predictive error for underestimating heart reserve function.
  7. 7. The method for monitoring cardiac functional recovery status of a population after coronary heart disease surgery according to claim 1, wherein the step 3 comprises: step 3.1, performing equal stress intensity pairing analysis; The pair of stress intensities must simultaneously satisfy the following conditions: the stress load matching comprises that the standard stress load factor difference value of the current time period and the historical pairing time period is less than or equal to 10%, the time window matching comprises that the time difference of the two time periods is less than or equal to 30 days, the circadian rhythm matching comprises that the circadian time difference of the two time periods is less than or equal to 2 hours, the data quality compliance comprises that the electrocardiosignal quality SQI of the two time periods is more than or equal to 60 minutes, and no data artifact exists; the requirements for early warning, triggering and judging of the stress intensity are as follows: The condition that the current CRI index is reduced by more than or equal to 5% and less than 10% relative to a historical baseline in a single effective pairing is marked as a rehabilitation decline trend to be observed, and the emotion-electrocardio response slope K is increased by more than or equal to 0.1 and less than 0.2 relative to the historical baseline in the single effective pairing; The method meets the following core conditions and triggers stress intensity early warning, wherein more than 80% of effective paired data in a continuous 2-week monitoring period show that the current CRI index is reduced by more than or equal to 10% relative to a historical baseline or the emotion-electrocardio response slope K is continuously increased by more than or equal to 0.2 in the continuous 2-week monitoring period; Triggering emergency high risk early warning, namely, in a continuous 3-week monitoring period, the accumulated drop amplitude of the CRI index is more than or equal to 20 percent, and the continuous abnormality of stress response characteristics is accompanied, wherein the CRI index is directly dropped to below 60 minutes under the same stress intensity, and is judged to be recovery decline; The corresponding treatment measures are as follows: The system automatically marks the time period data, brings in the current week trend tracking range, updates trend change every day, does not trigger early warning, and only prompts trend in the week rehabilitation report; Triggering stress intensity early warning, namely triggering secondary stroke risk early warning by a system, outputting a hidden recovery decay prompt report, and defining abnormal characteristics and potential risks; the following treatments are synchronously executed, namely, the daily activity intensity is recommended to be properly adjusted, the overfatigue is avoided, and the change of the resting state electrocardio characteristics is monitored every day; Triggering emergency high risk early warning, namely directly incorporating a three-level high risk comprehensive judging process and executing corresponding disposal measures; step 3.2, pairing analysis of the physiological states; The pairing of the physiological states must meet the following conditions simultaneously that the time periods are fixed and unified, namely deep rest time period data of 1:00-3:00 in the early morning are fixedly selected, the data quality is compliance, the signal quality SQI is more than or equal to 60 minutes, no physical disturbance and abnormal heart rhythm artifact are caused, the states are not stressed, namely no stress event is triggered in the corresponding time period, and the stress confidence coefficient outputted by the pre-stress identification model is less than 0.3; the physiological state early warning is a procedural monitoring condition, and the pathological degradation signal is a final pathological judgment standard; The physiological state early warning must meet the following conditions that the monitoring period completely meets the pairing admittance requirement, and in the corresponding period, at least 1 index is abnormal and deviates from a base line on a single day in the rest state QRS wave width, HRV low frequency power LF, LF/HF ratio, ST segment micro-offset and rest state heart rate 5 core indexes; The pathological degradation signal judgment standard condition can be met by calculating characteristic change trend through linear fitting based on depth resting state data for 14 continuous days, judging that any one of following persistent abnormality occurs, judging as a hidden recovery state degradation signal, and incorporating the two-stage stroke risk comprehensive judgment, wherein the resting state QRS wave width is continuously increased, the cumulative increase is more than or equal to 10%, the HRV low-frequency power LF is continuously increased, the cumulative increase is more than or equal to 20%, the LF/HF ratio is continuously unbalanced, the ST segment is continuously micro-offset, the cumulative offset is more than or equal to 50 mu V, the resting state heart rate is continuously increased, and the cumulative increase is more than or equal to 15%; The corresponding processing measures are that the system automatically marks abnormal indexes, tracks characteristic change trend every day, does not trigger grading early warning, only prompts the abnormal indexes and duration in a daily monitoring report, and reminds patients to take rest and avoid stress stimulation; The pathological degradation signal judgment comprises the steps of triggering a secondary stroke risk early warning by a system, outputting a hidden myocardial blood supply abnormality prompt report, and synchronously executing the following treatment, namely suggesting a user to consult a professional medical institution for further evaluation, synchronously pushing the abnormality report to a user rehabilitation management responsible person, and taking the abnormality report as an objective reference basis for rehabilitation follow-up evaluation; step 3.3, performing equal activity pairing analysis; The equal activity amount pairing must meet the following conditions that the activity types are matched, namely the activity types of the two time periods are low-intensity aerobic activities, the activity intensity is matched, namely the step speed difference value of the two time periods is less than or equal to 0.5km/h, the activity duration difference value is less than or equal to 10 minutes, the baseline state is matched, namely the patient is in a lying/resting rest state for more than or equal to 30 minutes before the two time periods are active, the data quality is in compliance, namely the electrocardiosignal quality SQI of the two time periods is more than or equal to 60 minutes, and no data is abnormal; the method comprises the steps of determining the initial screening triggering condition, determining the heart compensatory capacity early warning, namely, finishing characteristic comparison through activity pairing, triggering the activity early warning, and determining the heart compensatory capacity early warning if the activity is abnormal continuously; The early warning triggering judgment requirements are as follows: After effective pairing is completed, any characteristic abnormality occurs in the current period relative to the historical synchronous data, activity early warning is triggered, continuous tracking is started, wherein the heart rate recovery time after activity is prolonged by more than or equal to 15% and less than 30% under the same activity intensity, the ST-segment offset amplitude in activity is increased by more than or equal to 30 mu V and less than 50 mu V under the same activity intensity, and the R-wave amplitude in activity is reduced by more than or equal to 5% and less than 10% under the same activity intensity; the cardiac compensatory ability pre-warning conditions must be satisfied simultaneously with the following conditions: In a continuous 2-week monitoring period, more than 80% of effective paired data have at least 2 continuous anomalies, namely continuous prolongation of heart rate recovery time after activity is more than or equal to 30%, continuous increase of ST-segment deviation amplitude in activity is more than or equal to 50 mu V, continuous decrease of R-wave amplitude in activity is more than or equal to 10%, and CRI index decrease trend accompanied by stress intensity pairing is synchronized; The corresponding treatment measures are that the activity amount early warning is that the system automatically marks abnormal characteristics, incorporates daily trend tracking, prompts abnormal heart response after activities in daily monitoring reports, recommends patients to reduce activity intensity, shortens activity duration and observes symptom changes; The heart compensatory ability early warning comprises the steps of triggering a secondary stroke risk early warning by a system, outputting a heart compensatory ability decline prompt report, and synchronously executing the following treatments, namely suggesting a user to consult a professional medical institution for further evaluation, perfecting related examination, evaluating heart functions and exercise tolerance; and 3.4, comprehensively judging risk classification and performing closed loop treatment.
  8. 8. The method for monitoring cardiac functional recovery status of a population after coronary heart disease surgery according to claim 7, wherein the step 3.4 comprises: combining the paired analysis results of three dimensions, cross-verifying and distinguishing benign physiological fluctuation and rehabilitation state decline trend, outputting three-level grading risk early warning and corresponding closed-loop treatment measures, and synchronously storing all early warning results into a patient history database as a reference basis for baseline updating and model fine tuning; When the risk level is the first-level low risk, the comprehensive judgment standard is satisfied with any one of the following conditions: the CRI index is in a recovery stable region of 60-79 minutes, and has no continuous descending trend; Only transient electrocardiographic characteristic fluctuation occurs, and continuous abnormality is avoided for 3 days or more; none of the three dimension pairing analysis triggers early warning, and no clear pathological degradation signal exists; when the risk level is the risk in the second level, the comprehensive judgment standard is satisfied with any one of the following conditions: The pairing of the equal stress intensity triggers the early warning of the stress intensity, and the CRI continuously decreases by more than or equal to 10 percent but still more than or equal to 60 minutes for 2 weeks; The physiological state pairing is judged to be a pathological degradation signal, and continuous resting state characteristic abnormality occurs; The heart compensatory ability early warning is triggered by the pairing of the equal activity amounts, and the characteristic abnormality occurs after the continuous activity; the primary screening abnormality occurs in the pairing analysis of two or more dimensions, and the trend is continuously worsened; when the risk level is three-level high risk, the comprehensive judgment standard is satisfied with any one of the following conditions: CRI index <60 points, directly judging that the rehabilitation state is declined; An electric abnormality signal of the strict gravity center appears; the continuous 3-week drop of CRI is more than or equal to 20 percent, and the CRI is accompanied by abnormal heart function characteristics in two or more dimensions.
  9. 9. The heart function rehabilitation state monitoring system is characterized by comprising an electrocardiosignal preprocessing unit, wherein the electrocardiosignal preprocessing unit is connected with a multidimensional electrocardiosignal feature extraction unit, the multidimensional electrocardiosignal feature extraction unit is respectively connected with a third-order individuation baseline management unit and a prepositive stress identification AI unit, the third-order individuation baseline management unit and the prepositive stress identification AI unit are respectively connected with a core correlation deduction AI unit, and the third-order individuation baseline management unit, the prepositive stress identification AI unit and the core correlation deduction AI unit are respectively connected with an equal-condition longitudinal pairing early warning unit.
  10. 10. The heart function rehabilitation state monitoring system of the crowd after coronary heart disease operation according to claim 9, wherein the electrocardiosignal preprocessing unit, the multidimensional electrocardio feature extraction unit, the third-order individuation baseline management unit, the prepositive stress identification AI unit, the core correlation deduction AI unit and the equal condition longitudinal pairing early-warning unit are respectively connected with the data storage and calling unit; The electrocardiosignal preprocessing unit processes original electrocardiosignal data, filtering noise reduction, window division, quality assessment and time stamp synchronization are completed, qualified electrocardiosignal data are screened out, the multidimensional electrocardiosignal characteristic extraction unit processes the qualified electrocardiosignal data, three main types of core electrocardiosignal characteristics including heart rate reserve, stroke volume reserve and recovery capability are synchronously extracted to form a standardized characteristic stream, the third-order individuation baseline management unit and the prepositive stress identification AI unit process the standardized characteristic stream respectively, the third-order individuation baseline management unit completes baseline construction and characteristic deviation calculation according to the postoperative stage of a patient, the prepositive stress identification AI unit completes stress state identification and standardized stress load factor quantification, the core correlation deduction AI unit processes the standardized electrocardiosignal characteristics, the stress load factor and baseline envelope to complete stress-physiological coupling modeling, calculates and outputs real-time CRI of the patient, response slope K and recovery state classification, the condition longitudinal pairing unit processes the real-time CRI, the characteristic stream, the standardized stress load factor and the baseline data to complete three-dimensional pairing and early warning and decision and three-stage risk assessment and guide individual recovery trend analysis.

Description

Method and system for monitoring cardiac function rehabilitation state of crowd after coronary heart disease operation Technical Field The invention belongs to the technical field of health monitoring and physiological data processing, relates to a method for monitoring the heart function rehabilitation state of a crowd after coronary heart disease operation, and further relates to a system for monitoring the heart function rehabilitation state of the crowd after coronary heart disease operation. Background Coronary heart disease (coronaryheartdisease, CHD) is a disease of the cardiovascular system that results from coronary atherosclerosis causing stenosis or occlusion of the vascular lumen, resulting in ischemia and hypoxia of the heart muscle. The postoperative rehabilitation management is a key link for reducing the recurrence risk of restenosis, heart failure and malignant cardiovascular events in a bracket, clinical data show that the occurrence rate of coronary heart disease postoperative recovery failure reaches 3% -10% in 1 year, the occurrence rate of cardiovascular abnormal events exceeds 20% in 5 years after the operation, and the hidden heart reserve function decline exists before more than 80% of abnormal events occur, but the prior art outside-hospital monitoring means cannot effectively capture the early change signal. The heart reserve function refers to the maximum cardiac output lifting capacity of the heart relative to the resting state under the stress state, is a core quantitative index of the recovery state after coronary heart disease operation, the decline of the heart reserve function is an independent predictive factor of adverse cardiovascular events, and the heart recovery reserve index (Cardiac Rehabilitation Reserve Index, CRI) is a special index for quantifying the heart reserve function of a patient after coronary heart disease operation, has a value range of 0-100 minutes and has positive correlation with the recovery state and the heart compensation capacity of the patient. At present, the related technology mainly develops around the directions of coronary heart disease monitoring, postoperative management and the like, but no accurate evaluation scheme for adapting to the special requirements of postoperative rehabilitation is formed, and the prior art has the following core defects that the wearable electrocardiograph monitoring technology is characterized in that Chinese patent application with publication number CN107928651A and publication number 2018 and 4 month and 20 discloses a wearable coronary heart disease detection device, chinese patent application with publication number CN109758131A and publication number 2019 and 5 month and 17 discloses a wearable coronary heart disease detection device, and the technology acquires pulse waves through a PPG sensor (PhotoPlethysmoGraphy Sensor, photoelectric volume pulse wave tracing sensor) or a lead sheet, And (3) calculating basic physiological indexes such as heart rate, blood pressure, heart rate variability (HEART RATE Variability, HRV) and the like by using electrocardiosignals, and only realizing simple numerical overrun alarm. The method has the core limitations that a general population fixed threshold or a simple initial baseline is adopted, the pathological evolution rule of 'acute inflammatory phase-endothelialization phase-stabilization phase' after coronary heart disease operation is not matched, so that the baseline construction lacks medical logic support, meanwhile, the technology is mostly based on the static threshold for statistical judgment, the dynamic deviation of the electrocardio characteristic along with heart remodeling in the rehabilitation process is ignored, the judgment logic based on population statistics can not capture the fine evolution trend of the electrophysiological characteristic (such as QRS wave amplitude and ST segment deviation) of an individual in the rehabilitation process, so that serious hysteresis exists in early warning, only the instantaneous value of a single index is concerned, the correlation rule among the indexes is not mined, the heart reserve function can not be evaluated, and abnormal fluctuation of the rehabilitation state is difficult to capture. The digital coronary heart disease management technology is characterized in that a Chinese patent application with publication number of CN117524458A and publication number of 2024, 2 and 6 discloses a digital coronary heart disease management system, a Chinese patent application with publication number of 2023, 7 and 28 and publication number of CN121445388A discloses a coronary heart disease screening system based on exercise load electrocardio, and the technology collects medical history, physiological data and the like of a patient through a multi-dimensional information input platform and relies on an artificial intelligent model to identify risk factors and predict diseases. The method has the defects th